One prior art system for and method of automatically matching a drive circuit to a load circuit involves controlling series and shunt reactances of a matching network connected between the drive and load circuits. The reactances are controlled in response to indications of (a) the resistance seen looking into the matching network from the drive circuit and (b) phase relationships of the voltage and current supplied by the drive circuit to the matching network. The network is adjusted so that the impedance seen looking from the drive circuit into the matching network moves along a line of constant conductance or constant resistance as the value of a first of the reactances is varied. The first reactance is varied until the resistance looking into the matching network from the drive circuit equals the output impedance of the drive circuit, which is equal to the desired characteristic impedance seen looking into the matching network from the drive circuit. The network is then adjusted so that the resistance or conductance seen looking into the matching network from the drive circuit remains constant despite variations of the second reactance. The second reactance is varied until the current and voltage supplied by the drive circuit to the network have the same phase, at which time a match is assumed to exist and no further variations of the reactances of the matching network occur.
In the case of a high-Q load, i.e., a load having a reactive impedance component on the order of 100 times greater than its resistive component, it is difficult to generate control voltages precisely indicative of the resistance seen looking into the matching network from the drive circuit. The difficulty arises because of the inherent nature of circuitry utilized for detecting the resistance. In particular, resistance detectors include current and voltage sampling impedance elements across which are derived control voltages proportional to the current flowing in the matching network and the voltage applied to the matching network by the drive circuit. These control voltages are linearly combined to derive a resultant voltage that is indicative of the vector sum of the current and voltage samples. The resultant voltage is detected by a rectifying network which generates an output control voltage indicating the sensed resistance magnitude.
For a high-Q load, where there is a great possibility of an appreciable difference between the phases of the load voltage and current with respect to the relative phases of the corresponding voltage samples, errors aree likely to be induced in the control voltages because of unavoidable stray reactances in the sampling circuits. If the matching network is to operate over a relatively wide frequency range, such as two decades from 2 to 8 mHz, the strays cannot be compensated so there is a relatively high percentage of unavoidable error in the resistance indicating control signal.
Because of the strays, accurate resistance detection is not feasible, in practice, for load circuits having Q's greater than approximately 10. With the prior art resistance detecting technique, the ratio of the imaginary (reactance or susceptance) component to real (resistance or conductance) component becomes rather large when the value of the first reactance has been adjusted so that the resistance seen looking into the matching network from the drive circuit reaches the desired value. For example, if it is desired to match a 50 ohm driving circuit to a load having an impedance: EQU Z.sub.L = 0.1 + j24 1.
where:
Z.sub.L = the load impedance, and EQU j = .sqroot. - 1,
the impedance seen looking into the matching network is first transformed to: EQU Z.sub.A = 50 + j534.3 2.
where:
Z.sub.A = the impedance seen looking into the matching network from the drive circuit
after the first reactance has been adjusted and prior to adjustment of the second reactance. Therefore, to provide a proper match between the load and drive circuit, the 50 ohm resistance component must be detected accurately to properly control the adjustment of the first reactance of the matching network.
If a standard, prior art resistance detector, as discussed above, is used to sense the resistive component, the accuracy of detection is given by: EQU % error = (X.sub.A /R.sub.o) tan (.theta.) .times. 100 3.
where:
R.sub.o = the resistive component to be detected (R.sub.o = R.sub.A = 50 ohms in this example)
X.sub.A = Reactive component (X.sub.A = 534 ohms in this example)
.theta. = Relative phase angle between the current "I" and output voltage V.sub.I of the current sampling transformer.
Because of the stray reactance in the resistance detector, there is usually a phase deviation of one or two degrees between the voltage and current samples and the control voltages which are vectorially combined to form the resultant. For a phase deviation of .+-.1.degree. between one of the samples and the control voltage responsive to the sample, Equation (3) yields a resistance error of .+-.18 %. Obviously, such a large error cannot be tolerated in a matching network that is expected to automatically and accurately couple a load circuit to a drive circuit with maximum efficiency in a stable manner.
Another disadvantage with the prior art resistance detector when used with high-Q circuitry, is that the sensitivity of such a detector is degraded when a large impedance is detected. The sensitivity of a resistance detector is: ##EQU1## where:
E.sub.L = voltage across the load impedance
K.sub.I = current sample constant = (V.sub.I /I)
Z = magnitude of load impedance (Z = 536 ohms for the above example)
From Equation (4), it is seen that the sensitivity of the resistance circuit is inversely proportional to approximately 29,000, whereby an extremely low voltage is derived from the resistance detector for large load impedances.